View source on GitHub →

Issue #1001 Resolution Summary

Benchmark Results

Comprehensive benchmarks comparing trust_score() vs trust_score_detailed() performance.

Performance Comparison

Network Size trust_score() trust_score_detailed() Overhead
100 nodes 5.59 µs 11.68 ms ~2,090x
1000 nodes 5.75 µs 121.86 ms ~21,700x
5000 nodes 7.13 µs 617.00 ms ~86,500x

Bottleneck Analysis

The performance overhead breaks down as:

  1. Edge Collection (90%): get_all_known_dids() + filtering is O(N × M)
  2. Attestation Conversion (5%): TrustEdge → TrustAttestation mapping
  3. SHA-256 Hashing (5%): Actually quite fast

Key Insight: SHA-256 hashing is NOT the primary bottleneck. The real issue is the O(N × M) edge collection algorithm.

Hash Memoization Decision

Not Recommended

Hash memoization would only improve ~5% of total execution time. The real bottleneck is edge collection, not hashing.

Alternative: Implement a reverse edge index (target → sources) to reduce edge collection from O(N × M) to O(1).

Impact estimate:

  • With reverse index: 121 ms → ~6 ms (1000 nodes)
  • With hash memoization only: 121 ms → ~115 ms (minimal improvement)

block_in_place() Contention

Monitoring Thresholds

Defined acceptable thresholds for trust_oracle_block_in_place_total metric:

Status Threshold Action
Normal <100 calls/sec/core No action needed
Warning >500 calls/sec/core Audit callers, implement caching
Critical >1000 calls/sec/core Async refactor required

Async Refactoring

Making TrustService trait methods async is a future consideration:

  • Pro: Eliminates threadpool contention
  • Con: Breaking API change
  • Timeline: After Phase 6 crate consolidation (#861)

Documentation Added

  1. Inline Documentation

    • Added comprehensive doc comment to trust_score_detailed()
    • Includes benchmark results and use case guidance
    • Location: apps/trust/src/service_tokio.rs

  2. Performance Guide

    • Detailed bottleneck analysis
    • Optimization recommendations with tradeoffs
    • Monitoring and alerting guidance
    • Location: docs/performance/trust-service-performance.md

  3. Benchmark README

    • Instructions for running benchmarks
    • Expected results and interpretation
    • Troubleshooting guide
    • Location: apps/trust/benches/README.md

Recommendations

Immediate (Phase 22 - tier:3)

  1. ✅ Add benchmarks - Completed
  2. ✅ Document tradeoffs - Completed
  3. 🔲 Implement reverse edge index - New issue needed
  4. 🔲 Add query latency metrics - New issue needed

Future (Post-Phase 6)

  1. Async trait refactoring - After crate consolidation (#861)
  2. Hash memoization - NOT recommended (minimal impact)

Issue Status

Resolution: Issue #1001 objectives completed:

  • ✅ Benchmarks added and documented
  • ✅ Performance tradeoffs documented
  • ✅ Hash memoization evaluated (not recommended)
  • ✅ block_in_place() thresholds defined

Follow-up Issues:

  • Reverse edge index implementation (high impact optimization)
  • Query latency metrics (observability improvement)

Related:

  • Issue #877: Multi-graph trust implementation
  • PR #987: trust_score_detailed() implementation
  • Issue #861: Phase 6 crate consolidation
← Back to Docs